RE: less-restrictive range and domain terms

Hi Stephen,

Rhoads, Stephen writes:
 > I have recently had a similar (although not precisely the same) problem with
 > the UI of Protege.
 > 
 > I particular, if you:
 > 
 > (a) Choose not to use global domain and range constraints (for, among other
 > reasons, the "open world" assumption), and
 > 
 > (b) Have lots of optional and/or datatype properties which do not make sense
 > to include in DLs for defined classes,
 > 
 > then these properties will not show up at the (intended) classes in
 > individuals forms.  One must explicitly modify the forms for classes and the
 > resultant data is stored in a separate file from the OWL ontology.
 > 
 > If someone is looking at the ontology for the first time (or is otherwise
 > not an expert in the domain of discourse) it is not immediately clear which
 > classes are intended to be used with these types of properties.
 > 
 > It would almost make sense to introduce a couple of annotation properties to
 > say, for example, "this property is intended to be used with the following
 > class (or classes), or "this property is intended to take as values the
 > following class (or classes).  This information would then be stored in the
 > OWL file and could be directly exploited by ontology editors to offer up
 > reasonable options to users.
 > 
 > Just a thought.
 > 
 > --- Stephen

That's pretty much the UI problem I'm trying to address. Also, authors
of rdf schemas often don't want to specify a strong typed rdfs:Range
for fear of closing down the utility of the schema, and so set the
rdfs:range to be of rdfs:Resource. It would be nice to be able to hint
at the sort of classes that were intended to be the range of these
properties without completely restricting them globally.

I suspect (but don't know) that this is the reason for rdfs:Resource
ranges in e.g. the foaf schema[1] (see currentProject, depiction, mbox
etc..)

Cheers,

Phil

[1] http://xmlns.com/foaf/0.1/

Received on Tuesday, 4 May 2004 15:59:36 UTC